Tajti Attila, Szalay Péter G
Institute of Chemistry, Eötvös Loránd University , Budapest H-1125, Hungary.
J Chem Theory Comput. 2016 Nov 8;12(11):5477-5482. doi: 10.1021/acs.jctc.6b00723. Epub 2016 Oct 17.
Describing electronically excited states of molecules accurately poses a challenging problem for theoretical methods. Popular second order techniques like Linear Response CC2 (CC2-LR), Partitioned Equation-of-Motion MBPT(2) (P-EOM-MBPT(2)), or Equation-of-Motion CCSD(2) (EOM-CCSD(2)) often produce results that are controversial and are ill-balanced with their accuracy on valence and Rydberg type states. In this study, we connect the theory of these methods and, to investigate the origin of their different behavior, establish a series of intermediate variants. The accuracy of these on excitation energies of singlet valence and Rydberg electronic states is benchmarked on a large sample against high-accuracy Linear Response CC3 references. The results reveal the role of individual terms of the second order similarity transformed Hamiltonian, and the reason for the bad performance of CC2-LR in the description of Rydberg states. We also clarify the importance of the T̂ transformation employed in the CC2 procedure, which is found to be very small for vertical excitation energies.
准确描述分子的电子激发态对理论方法来说是一个具有挑战性的问题。像线性响应CC2（CC2-LR）、分区运动方程多体微扰理论（2）（P-EOM-MBPT(2)）或运动方程耦合簇单双激发二阶微扰理论（EOM-CCSD(2)）等流行的二阶技术，其结果往往存在争议，并且在价态和里德堡态上的准确性也不均衡。在本研究中，我们将这些方法的理论联系起来，并为研究它们不同行为的起源建立了一系列中间变体。这些变体在单重价态和里德堡电子态激发能上的准确性，在一个大样本上以高精度线性响应CC3为参考进行了基准测试。结果揭示了二阶相似变换哈密顿量中各个项的作用，以及CC2-LR在描述里德堡态时表现不佳的原因。我们还阐明了CC2程序中使用的T̂变换的重要性，发现它对于垂直激发能非常小。
